Int J Advanced Design and Manufacturing Technology, Vol. 7/ No.

3/ September - 2014 11

Optimization of Suspension
System of Sport Car in Three
Dimensional Reactions Space
R. Kazemi*
Department of Mechanical Engineering,
K. N. Toosi University of Technology
E-mail: kazemi@kntu.ac.ir
*Corresponding author

D
M. Raf’at & H. Zolghadr
Department of Mechanical Engineering,
K. N. Toosi University of Technology
E-mail: fgh.rafat@gmail.com, sdsad9079@yahoo.com

S I
Received: 18 January 2014, Revised: 25 February 2014, Accepted: 27 Jun 2014

Abstract: Suspension system as one of the most important key elements of vehicle

f
has always been considered by research centres and auto makers over the world.
This is of higher importance when it comes to sports cars which work within

o
especially sensitive conditions and their performances may be affected adversely
even by the smallest defects. This study investigates optimization of effective
handling parameters of a sport car with three optimization indices covering
racetrack at the shortest possible time. For this purpose, a sport car model

e
simulated in ADAMS/Car software has been used and all test steps have been
implemented in this environment. To reach optimum solution in different racetrack

v
conditions, Pareto solution set optimization method was used in which

i
optimization indices have been examined covering distance in standard constant
radius test with three different radii. Implementing these experiments at different
levels of optimization parameters, all solution sets were collected in a three

h
dimensional diagram called performance space whose coordinates axes are
optimization indices values; then using LP norm idea, possible optimum solutions

c
were selected among other ones. In order to find optimum solution in a special
racetrack, a new idea was proposed through which weighted coefficients related to

r
optimization indices for a special racetrack were determined and optimum
solutions appropriate for that track were selected from Pareto solution set.

A
Keywords: Optimization, Pareto Solution Set Method, Three Dimensional
Reactions Space, Sport Car

Reference: Kazemi, R., Raf’at, M., and Zolghadr, H., “Optimization of

Suspension System of Sport Car in Three Dimensional Reactions Space”, Int J of
Advanced Design and Manufacturing Technology, Vol. 7/ No. 3, 2014, pp. 11-18.
Biographical notes: R. Kazemi received his BSc and MSc in mechanical
engineering from Isfahan University of Technology (IUT), Iran, in 1993 and 1995,
respectively, and his PhD in mechanical engineering from Amirkabir University of
Technology, Iran, in 2001. He is an Associated Professor of Mechanical Engineering
and head of the Advanced Vehicle Systems Research Center (AVSRC) at K.N.Toosi
Univ. of Technology, Tehran, Iran. He is currently involved in modelling, design, and
improvement of vehicle steering, suspension, and braking systems and also Vehicle
Dynamics Control systems. M. Raf’at received his BSc and MSc in mechanical
engineering from Semnan University and K.N.Toosi University of Technology
(KNTU) Iran, in 2010 and 2013, respectively. He is interested in Vehicle
Subsystem Design and Optimization, Vehicle Dynamic Control and Vehicle Safety
System Design. H. Zolghadr received his BSc and MSc in mechanical engineering
from K.N.Toosi University of Technology (KNTU) Iran, in 2005 and 2008,
respectively. He is interested in nonlinear control and adaptive control for non
predictable systems.

www.SID.ir
© 2014 IAU, Majlesi Branch
12 Int J Advanced Design and Manufacturing Technology, Vol. 7/ No. 3/ September– 2014
1 INTRODUCTION
General vehicle design, its tuning and also set up in
different racetracks are up to engineering department of
each team which provides optimum conditions for
driver to cover racetrack faster than others. Any small
change in vehicle basic design or at its set up time may
contribute to increase vehicle efficiency and in the case
of capable drivers, results in increased efficiency in
traversing racetrack [1], [2]. In addition to general
vehicle designing, computer softwares have been used
as well for setting up simulation and race car
optimization in different racetrack conditions. A sport
car model simulated by ADAMS/Car software has been
used in the current study to implement the relevant tests.
Vehicle designing is a significant example of
optimization designing problems in which different
characteristics such as different dynamic and
aerodynamic features, tires, engine, shock absorber,
driver’s decision, etc. may affect vehicle controlling
approach and efficiency.
This research is aimed at finding optimum solutions for
vehicle dynamic parameters subjected to different
racetracks in various tracks. Along a racetrack, the
driver faces with turns with different radii and also
straight paths. Each turn requires related optimum
characteristics of the vehicle. To cross total racetrack,
thus, it is necessary to provide optimum conditions for
v
all the mentioned settings [3]. In order to optimize
i
sensitive parameters of suspension system that is
previously determined by sensitive analysis method [4],
h
handling characteristics of a sport car have been
optimized in this paper through Pareto solutions set. A
c
set of appropriate solutions were collected and via a
method explained later on optimum solution selection,
r
optimum indices of suspension system have been
determined in different racetrack conditions.
Fig.1 used parameters of front suspension
© 2014 IAU, Majlesi Branch
2 OPTIMIZATION PARAMETERS
A sport sedan was considered in this study to be
converted to a race car. For this purpose, some
optimizations were implemented on its suspension
system. Its front suspension system of Mcpherson type
associated with anti roll bar and its back suspension
system is multi link type with five arms and back anti
roll bar. Good capability to improve journey comfort
and handling are advantages of this system [5]. Also, its
steering system is of rack and pinion type and other
modelled elements such as tires, sub-chassis and body,
engine and powertrain system are simulated according
to original car characteristics.
D
Optimization parameters in front and back suspension
system of the sample car, a tuned sedan simulated by
S I
ADAMS/Car software, are shown in Figs. 1 and 2. The
mentioned parameters with applied changes during
optimization are available in Table 1.
3
o f
OPTIMIZATION INDICES
In order to analyze vehicle handling, many indices can
be examined, including maximum roll angle, roll angle
value (subgraph area), roll angle value in static status,
e
lateral acceleration lunge, rising time of lateral
acceleration, lateral acceleration value in static status,
yaw rate lunge, rising time of yaw rate, yaw rate value
in static status and time differences between 50% of
steering wheel angle value and 50% of static status of
lateral acceleration and also in frequency rang indices
such as yam rate RMS and lateral acceleration RMS.
These characteristics determine vehicle handling quality
through affecting temporal solution area and system
static status [6], [7].
Fig. 2 Back suspension parameters used
www.SID.ir
Int J Advanced Design and Manufacturing Technology, Vol. 7/ No. 3/ September - 2014 13

Table 1 Optimization parameters

Level Level Level Level Level levels
-2 -1 0 1 2 parameters

-3 -2 -1 0 1 A (degree)
108 114 120 126 132 B (Nm/deg)
0.7 0.8 0.9 1 1.1 C (coefficient)

Table 2 Analysis characteristics

Analysis Turning radius Experiment time Initial velocity Final velocity
characteristics (m) period (s) (km/h) (km/h)
Analyze I 30 11.5 35 60
Analyze II 80 18.5 60 90

D
Analyze III 160 22.25 90 140

An index must be chosen to fulfil the goal of this

research that is to optimize a sport car and the concept
of "velocity race", indicating traversing a special
distance at the shortest possible time as the most
S I
4 PARETO SOLUTION SET OPTIMIZATION

f
METHOD
important parameter to compare these vehicles.
Therefore the index of "traversed distance value at a Pareto solution set have been used as a border line of

o
constant time along a special course" has been selected efficient solution in performance space regarding the
which indicates required time to cover a racetrack. In optimization problems of indices proposed in this
order to cover this index of all the above mentioned research. Advantage of using this method is to achieve

e
characteristics related to vehicle handling, the selected a series of solutions; each of its constituents may be
track for analyzing a standard track has different curves counted as an optimum solution in a special racetrack
that a vehicle with certain radius turns inside it.

i v
Assessing vehicle reaction in different radii needs to be
carried out in three different situations reflecting low,
intermediate, and high limit racetrack radii. Since
condition. So, in facing with different racetracks,
optimum parameters can be determined among few
options via a new method which will be explained later.

h
Similar to all selected methods, this one consists of two
racetrack is designed according to the type and basic steps: potential solutions determination and

c
performance of participating vehicles, selected radii for selection from among them [9].
this research experiments are calculated by considering In Pareto analysis method, there are rules and

r
wheel steering roll limitation on one hand, and
calculating maximum available velocity and lateral
acceleration that the vehicle can bear on the other hand
were considered.
conditions for a solution to be included in Pareto
selected solution series which must be fulfilled. These
conditions are mathematically expressed as in Eq. (1).

A
[8]. The minimum and maximum radii, accordingly,

An optimization index in Pareto method was adopted

for constant radius experiments of 30, 80, and 160m
and traversed distance value in each. Analysis index (І)
with 30m radius was considered as X axis index, and
analysis indices of (П) with 80m radius and (III) with
A x* design vector is a Pareto optimum point if and
only if for each x and I, Eq. (1) exists [3]:

( ) ( )
( ) ( )
(1)

160m radius as Y and Z axis indices respectively. The

mentioned analyses characteristics are shown in table 2.
Tests of this analysis are implemented in ADAMS/Car
software and by driver control file (DCF) - driven
constant radius cornering analytic method. This is a
standard test to specify vehicle handling characteristics
in stable status.
4.1. Pareto reactions space
Pareto solution sets are usually analyzed in a
performance space which by considering designing
indices is shown as coordinate axes, and includes
performance of the system under study in all possible
designs. To this aim, a point has been drawn in each
design, which is related to a certain value in any index.
After selecting optimization parameters and their
variable levels, the vehicle optimization solutions are
separately obtained for each index through analyzing

www.SID.ir
© 2014 IAU, Majlesi Branch
14 Int J Advanced Design and Manufacturing Technology, Vol. 7/ No. 3/ September– 2014

constant radius at 30, 80 and 160m that for each
analysis, the combination of all parameters levels must
be examined. Because the number of parameters are 3
and each of them has 5 levels, for each analysis ((53)
=125) test must be conducted. Therefore, by
implementing 375 testes for 3 indices, 125 points can
be drawn in performance space that its longitudinal,
latitudinal, and vertical positions, all determined by
traversed distance, are demonstrated in (І), (Π), and
(III) analyses for each combination of parameters.
Pperformance space, thus, has been obtained from
optimization parameters as shown in Fig. 3.

I D
f S
e o
i v
c h
Fig. 3 Pareto performance space and optimum solution in common weighted indices condition

Table 3 Parameters characteristics related to Pareto solution set

Number of
experiments
Camber angle
of front wheel (deg)
Rear Anti-roll bar
stiffness (Nm/deg)
Stiffness coefficient A
28

-3

114
r 103

-3

132
3

-3

108
61

-1

120
70

-3

114
2

-3

108
26

-3

114
10

-2

108
101

-3

132

of front and rear 0.9 0.9 1 0.7 0.8 0.8 0.7 1.1 0.7
springs
Traversed distance in
analyze I (mm) 137220 137214 137225 137227 137160 137161 137100 137230 136944
Traversed distance in
analyze II (mm) 639534 639535 639507 639498 639558 639547 639572 639422 639588

4.2. Pareto solution set experiment, the research is aimed at maximizing

Pareto solution set property is that for each point in the
mention set there is no corresponding point in
performance space which simultaneously has better
performance in both indices axes [3]. Since the
considered index covers distance value during the
indices of this problem; therefore, for selected points to
be included in Pareto solution set, there must be no
point with simultaneously greater value in both axes.
Table (3) shows parameters value related to Pareto
solution set.

© 2014 IAU, Majlesi Branch www.SID.ir

Int J Advanced Design and Manufacturing Technology, Vol. 7/ No. 3/ September - 2014
In Fig. 3, with regard to weighted coefficients of each
index considered as 1, covered distance value in each
point of performance space is equal to sum of covered
distance in each related analysis of that solution.
4.3. Optimum solutions selection
LP norms is one the most applied methods in Pareto
solution set to determine an optimum point. In this
method that was developed by Eschenauer in 1990,
minimum distance to Pareto solution set with respect to
an ideal solution (ideal point) to get optimum solution
is obtained as Eq. (2) [6]:
(∑ ( ( ) ) )1/p
(2)
Current applications of LP norms are L1, L2, L∞ where
P is equal to 1, 2, and infinite (∞) respectively.
This approach and L2 norms (p=2) are also used
similarly in this research to provide an appropriate
method to find required solution to optimize vehicle
handling performance when covering racetrack under
all different track conditions. The Optimum solution is
a combination of optimization parameters that by using
them, the vehicle could traverse the most possible
distance in a special course.
Using distance functions such as L2 norms makes the
concept of ideal point very important, because this
v
point is the best one to be obtained theoretically.
i
According to L2 norm method, optimum point has the
shortest geomantic distance to the ideal in Pareto
solution set, where this distance is calculated as a
h
vector in performance space.
To select optimum solution from among Pareto
c
solution set as mentioned above, L2 norm method was
r
used to select the ideal point. Since this research
indices cover distance value in constant radius
experiment, infinite point must be selected as the
A
theoretical ideal point for both indices. However,
because this point is geometrically meaningless in
performance space, it is impossible to use it. Thus, a
new suggested way is selecting zero point as the worst
possible solution for all three indices. Thus, the
optimum point would be a spot with uttermost distance
from this point. The ideal point based on Pareto
solution set is determined by drawing a sphere with
Pareto performance space coordinates origin as its
centre; the sphere radius increases and thus, the last
spot which sphere crosses would be the optimum point.
The optimum point solution obtained by this method
with both indices having equally weighted coefficients
is shown in Fig. 3.
In this section, indices with significant weighted values
are assumed to be fixed and equal to 1. In other words,
as observed in Fig. 3, both indices have the same
certain relative significance; here solution number 3 is
15
counted as the optimum solution. The performance
space transformation due to significance change of
each designing index will be discussed in the
paragraphs to come.
4.4. Scaling performance space axes
Assuming Fig. 3 has been drawn by considering
significance equal to 1 for both indices. Now if analysis
(І) index significance is multiplied by 2, that means x
axis index significance would be twice z and y axes
indices; then related diagram would be elongated on x
axis direction. Therefore, diagram shape varies and
possibly another design point would be considered as
the optimum solution.
D
It is worthy of note that the existing design point in
Pareto solution set would not differ with respect to the
I
prior case. In other words, Pareto solution set points are
independent of significance value of design indices and
S
the only difference is their drawing approach in
performance space. Also in this stage, use of an
accurate weighted ratio has no significance, and the
f
important issue in this connection is the quality of
performance space transformation due to relative
o
significance alteration of indices weight.
The objective of this section is to determine required
weighted coefficient for each Pareto performance space
e
indices calculated considering racetrack conditions.
Since each index indicates vehicle behaviour in a
certain turn radius, the importance of these indices
depends on using turns radii in different racetrack.
As seen in Fig. 4, the racetrack, as an example of a
course in the US, is a composition of turns with
different radii and also straight paths. Vehicle optimum
composition in a turn with radius of 160m observed in
right corner of path is different from another one with
radius of 30m observed in the centre of picture. So, to
design a vehicle with the best performance in all turns
of a racetrack, optimum selection from among all
existing parameters is needed.
Fig. 4 An example of a racetrack
In this step in order to introduce each index
significance effect, its axis must be multiplied by a
www.SID.ir
© 2014 IAU, Majlesi Branch
16 Int J Advanced Design and Manufacturing Technology, Vol. 7/ No. 3/ September– 2014
weighted coefficient to get the appropriate performance
space of this special problem. Here, the goal is to
provide an appropriate method to calculate the
abovementioned coefficient. As seen in Table 2,
experiments radii are selected so that vehicle velocity
in these 3 radii includes all vehicle efficiency to
traverse racetrack. So, the vehicle crosses a turn with
radius of 30m by ultimate velocity of 60 km/h and a
turn with 80m radius by 60 to 90 km/h velocity and a
turn with 160m radius by 90 to 140 km/h. Thus,
racetrack different radii considering maximum velocity
by which a vehicle could cross, could be assigned to
one of 3 intervals and be considered equal to one of
source radii. Thus, the racetrack is assumed to have
turns just with 30, 80, 180 m radii. Moreover, weighted
coefficient of each index would be determined by
distance value a vehicle traverses in each turn. In other
words, in addition to the number of turns, another
factor indicating course length of each turn curve
affects indices with significant value.
Since constant radius experiments are implemented in
approximately the same turn numbers, covered course
length in each experiment, therefore, it is different from
o
others and depends on curve radius (L=rƟ ). For this
v e
h i
r c
A
Fig. 5 Optimum point determination with weighted coefficient 5 fold to x axis
© 2014 IAU, Majlesi Branch
reason, since in experiments course length effect on
turn is introduced with respect to its radius, to introduce
turn length effect during the racetrack, an angle size is
used in which the turn rolls with constant radius. So,
the index weighted coefficient related to each turn is
equal to turn angle value. Multiplying each turn
weighted percent by its angle value and summing up
these values for all turns, total coefficient related to
each index is calculated. To simplify the calculation,
"π" can be eliminated from coefficients of all turns and
to apply changes upon just one of performance space
axis, higher coefficient can be divided by the lower,
and the resulted coefficient will be applied on the
related axis. So, applying indices weighted coefficients,
D
optimum solution may be obtained by using L2 norm
method as discussed in previous section.
I
As an instance of the resulted changes in the Pareto
solution set and also obtained optimum solutions due to
S
changes in optimization indices weighted coefficient,
the rest part focuses on the optimum solution finding
given that each weighted coefficient is 5 fold to other
f
indices. For each case, the diagrams as shown in Figs 5,
6 and 7 are obtained.
www.SID.ir
Int J Advanced Design and Manufacturing Technology, Vol. 7/ No. 3/ September - 2014 17

I D
Fig. 6

f S
Optimum point determination with weighted coefficient 5 fold to y axis

e o
i v
c h
A r
Fig. 7 Optimum point determination with weighted coefficient 5 fold to z axis

Fig. 5 indicates Pareto solution set in such a status that
x axis is multiplied by 5 and Fig. 6 shows Pareto
solution set while y axis is multiplied by 5 and Fig. 7
shows it when z axis index is multiplied by 5. In these
three diagrams, as seen axis related to indices with
weighted coefficients, Pareto solution set is exposed to
elongation and shape of Pareto solution set in these
diagrams are different from each other as well as Fig. 4.
Therefore, sphere showing optimum solution will pass
different points on Pareto solution set, as obtained
optimum points in the first, second and third statuses
are 3, 3 and 70 respectively.
5 CONCLUSION
This paper attempted to provide an appropriate method
which racing teams could be able to prepare their

www.SID.ir
© 2014 IAU, Majlesi Branch
18 Int J Advanced Design and Manufacturing Technology, Vol. 7/ No. 3/ September– 2014
vehicles for different races at the best situation relevant
to each racetrack, and make drivers capable of covering
racetracks in the shortest possible time via a simulated
model of sport car in ADAMS/Car software.
Accordingly, after determination of vehicle suspension
system sensitive parameters which have great influence
on desired handling indices, Pareto optimization
method was elaborated. Considering advantage of this
optimization method including capability of optimizing
more than one index and also a semi-optimum solution
set for different conditions, its each member could be
an optimum solution in a special condition of different
racetracks, where the method is adopted to achieve
sport car optimum suspension system.
In this research and according to problem demanded
solution type which is vehicle optimum parameters
determination to cover racetrack in the shortest possible
time, "covered distance value in a certain course with
constant radices" as our index was used. So, the
standard constant radius experiment implementation in
30, 80 and 160m radii that includes all velocity ranges
a racetrack considering discussed vehicle efficiency
range, provides three optimization indices.
By carrying out 375 experiments, obtained values are
drawn in Pareto performance space diagram and then
by using L2 norms method, a set as optimum solution
was selected. Considering different racetrack
conditions and race turn characteristics, a certain
v
weighted coefficient was used for each index. This
i
weighted coefficient is determined considering turn
radius and its roll angle and use of explained method,
then it is multiplied by performance space. Hence,
h
solution set form changes according to racetrack
condition and therefore, the optimum solution related to
REFERENCES
c
each automotive road race is determined.
r
[1]
A
Nerg, J., Rilla, M., Ruuskanen, V., Pyrhonen, J., and
Ruotsalainen, S., “Design of direct-driven permanent
© 2014 IAU, Majlesi Branch
magnet synchronous motors for an electric sports
car”, International Conference on Electrical
Machines, ICEM 2012.
[2] Kasprzak, E., Gentz, D., “The Formula SAE Tire Test
Consortium-Tire Testing and Data handling”, SAE
Technical Paper 2006-01-3606, 2006, doi:
10.4271/2006-01-3606.
[3] Kasprzak, E. M., Lewis, K. E., “Pareto analysis in
multi objective optimization using the co linearity
theorem and scaling method”, Motorsports
Engineering Conference Proceedings, Vol. 1: Vehicle
Design and Safety, 1998, pp. 340/1.
[4] Jiang L.-b., Liu J.-l., “Vehicle dynamics simulation
and influencing factors analysis of mini sporty car”,
International Conference on Electronics,
Communications and Control (ICECC), IEEE 2011.
[5] Nuti, A., Orives, R., Borini, G., and Filho, O.,
D
"Chevrolet montana - from a road car to a racecar",
I
SAE Technical Paper: 2004-01-3372, 2004, doi:
10.4271/2004-01-3372.
[6] Kazemi, R., Shariatpanahi, M., Salim, Sh., and
S
Ansari, Mm., “Mcpherson suspension system
geometry optimization based on vehicle directional
properties improvement by using biological pattern”,
f
Vehicle Dynamic General Department, Iran khodro
Search Center, 2000.
[7] Kazemi, R., Meidani, M., “Dynamic sensitivity
o
analysis of passenger car”, MSc. dissertation, K. N.
Toosi University of Technology, Mechanic faculty,
2005.
e
[8] Barkman, A., Tan, K., McIntosh, A., Hylton, P., and
Otoupal-Hylton, W., "Examination of an alternative
suspension configuration for a sprint car", SAE, 2014
World Congress & Exhibition.
[9] Siegmund, F., A. H. C. Ng., Deb, K., “Finding a
preferred diverse set of pareto-optimal solutions for a
limited number of function calls”, Congress on
Evolutionary Computation (CEC), IEEE, 2012.
[10] Gillespie, T. D., “Fundamentals of vehicle
[11]
[12]
dynamics”, SAE, ISBN: 1-56091-119-9, 1992.
Dixon, C. John., “Tires, suspension and handling”,
Second Edition, SAE, ISBN: 1-56091-831-4,
Cambridge University Press, U.K., 2007.
Milliken, W. F., Milliken, D. L., “Race car vehicle
dynamics”, SAE, ISBN: 1-56091-526-9, 1995.
www.SID.ir